(Bloomberg) – In a week when heavy rains in Louisiana led to eight deathsand forced more than 20,000 people from their homes, theU.S. said it can now tap a supercomputer to betterforecast flood events across all of the nation's waterways.

The new tool will let forecasters target potentially life andproperty threatening downpours street by street, giving people moretime to get out of harm's way or prepare for the worst if theycan't, said Thomas Graziano, director of NOAA'sOffice of Water Prediction.

'Quantum leap forward'

"It's NOAA's first foray into high-performance computing forwater prediction," Graziano said in a telephone interview. "It'sgoing to provide information in all kinds of areas where we havenone today. It's a quantum leap forward."

The new computer model will draw data from more than 8,000U.S. Geological Survey sites to simulate 2.7million locations in the 48 contiguous states, generating hourlyriver and stream forecasts, the National Oceanic andAtmospheric Administration said in a statement Tuesday. Earliercomputer models only used 4,000 sites every few hours.

With the new system, forecasters can pinpoint floods as much asthree days in advance, Graziano said. The model, when used withother forecasting systems, should be able to look out 10 to 30 daysfor potential storms that could cause flooding.

Predict potential floods days in advance

Since Aug. 8, 20.13 inches (51 centimeters) of rain has fallenon Baton Rouge, according to the National Weather Service, pushing rivers, streams andbayous to major flood levels. More than 20,000 people have beenforced from their homes by the rains, which come on the heals ofsevere floods in Texas and West Virginia earlier this year.

"There are many parts of this country where we are hydro blind,"Graziano said.

The goal is to be able to predict potential floods days inadvance. The model was developed along with the National Centers forAtmospheric Research, the Consortium ofUniversities for the Advancement of Hydrologic Sciences and theNational Science Foundation, among others.